Pith. sign in

REVIEW 5 major objections 6 minor 53 references

MolMetaLM: a Physicochemical Knowledge-Guided Molecular Meta Language Model

T0 review · 5 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read MolMetaLM claims that pretraining a SMILES-only decoder on structured physicochemical knowledge triples yields one model for generation, prediction, conformation, and optimization.

desk verdict Novel meta-language pretraining idea with open code, but the headline results are at risk from likely PubChem overlap and missing error bars. read the letter →

arxiv 2411.15500 v1 pith:OCVV67ZA submitted 2024-11-23 cs.ET cs.CL

Yifan Wu , Min Zeng , Yang Li , Yang Zhang , Min Li This is my paper · ORCID
classification cs.ETcs.CL
keywords molecularmetalanguagephysicochemicalknowledgedenoisingpretrainingproperty-conditionedgenerationpropertypredictionconformationinferenceoptimizationSMILES
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to show that a single molecular language model can cover property prediction, molecule generation, conformation inference, and molecular optimization without task-specific architectures or natural-language grammar. The proposal is to write each molecule together with its physicochemical properties as a set of knowledge triples $\langle S, P, O\rangle$ that share the same subject $S$ (the molecule's SMILES string), and to pretrain a transformer decoder to repair three kinds of noise in those sequences. Because hundreds of properties and fingerprint/conformation tasks can be combined, the scheme generates tens of thousands of pretraining tasks from one corpus. A sympathetic reader would care because, if this works, one SMILES-only pretrained model replaces several specialized models and unifies task formats.

What carries the argument

The load-bearing object is the molecular meta language sequence: a linear string $s_1,\dots,s_l, p_1, v_1,\dots,p_k,v_k$ that packs a SMILES molecule and $k$ property name–value pairs into one input, viewed as $k$ triples $\langle S, P_i, O_i \rangle$ sharing the same subject. Pretraining is a denoising scheme over this format: token-level noise (mask/replace tokens) teaches structure; sequence-level noise (delete or blank whole parts) teaches conditional generation; order-level noise (shuffle) forces the model to reconstruct a valid SMILES and re-pair property names with values. From more than 400 properties, five fingerprint types, and a translation/rotation-invariant local-coordinate encoding of conformations, the authors generate 18 basic denoising task types that expand to tens of thousands of pretraining tasks. This mechanism is what unifies discriminative and generative abilities in one decoder.

What would settle it

Take a single property, such as molecular weight, and condition generation on a dense grid of held-out numeric values spanning several orders of magnitude. If the correlation between conditioned values and generated molecules' computed values collapses to near zero or the model falls into repetitive loops on those held-out ranges, the claim that the meta language captures numerical semantics would be falsified. A simpler check: retrain with all property values replaced by random tokens; if property-conditioned generation and property prediction remain at the same accuracy, the property values themselves are not load-bearing.

Watch

Extended reading notes

Core claim

The central discovery claimed is that a molecule-specialized meta language—formatted as $\langle S, P, O\rangle$ triples with the molecule as shared subject, properties or fingerprints or conformations as objects—lets a decoder-only language model learn physicochemical semantics directly from text. After pretraining on over 110 million SMILES strings with 402 cheminformatics descriptor properties plus fingerprint and conformation prediction tasks, the paper reports proficiency across the four task families. For generation, MolMetaLM matches or beats text-based molecular language models on single- and multiple-condition generation and achieves high valid, unique, and successful ratios. For prediction, it outperforms dedicated models on most regression and classification benchmarks, and it exceeds scoring-function docking baselines on a blind-docking test once the RMSD cutoff passes 1.4 Å. The authors interpret the mechanism as building a SMILES–physicochemical-knowledge embedding space in which the model retrieves memorized triples and performs analogical reasoning.

Load-bearing premise

The scheme assumes that writing a continuous number as text tokens preserves enough of its meaning—order, scale, and distance—for the language model to learn and condition on it; the paper itself shows failures on properties like Chi3v, Kappa1, and Ipc when that meaning is lost.

Editorial extensions

If this is right

  • A single pretrained decoder can be fine-tuned for molecular property prediction by pooling its token representations, and on most of the tested regression and classification benchmarks it beats task-specific models trained for those datasets.
  • Property-conditioned generation from text constraints works better when the model is pretrained on property triples than when it is pretrained on natural language, because the model embeds numerical constraints with higher sensitivity.
  • Fingerprint and conformation information can be treated as ordinary property values in the same triple format, so generated molecules can be steered by fingerprint similarity or by reference backbone coordinates without changing the model.
  • Triple-format pretraining makes molecular representations more linearly separable on binary classification datasets, suggesting the physicochemical knowledge is embedded rather than merely memorized at the token level.
  • Because the format is extensible, new molecular knowledge can be added as new predicates $P$ with corresponding values, so the framework could absorb tasks beyond the four evaluated families without architectural changes.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper stops short of testing a non-discretized continuous embedding for property values; if the token bottleneck is real, such an embedding should improve generation for rare and extreme values.
  • The repetition failure on unseen numeric constraints raises a hypothesis the paper does not separate: some generation may be interpolation over memorized triples rather than computation over value semantics; holding out whole value ranges during pretraining would settle it.
  • The analogical-reasoning view suggests an interpretability tool the authors did not build: retrieve the nearest memorized $\langle S,P,O\rangle$ triples for a generated molecule and display the property context that drove the output.
  • Because the only input required is a SMILES string, the framework could in principle scale to any large compound collection with properties computed on the fly; the paper leaves this scaling direction implicit.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 6 minor

Summary. The paper proposes MolMetaLM, a decoder-only molecular language model pretrained on structured sequences of the form <S,P,O> (SMILES, property name, property value) to encode physicochemical knowledge. Pretraining uses denoising objectives at token, sequence, and order levels over a large corpus of PubChem SMILES (~110M molecules) with 402 RDKit descriptors, plus fingerprint and conformation prediction tasks. The authors evaluate the resulting model on molecule generation (single- and multi-condition, unconditional, and optimization), property prediction (AGBT regression, MoleculeNet classification), GPCR activity prediction, and blind docking on CASF-2016, and they claim broad proficiency across these tasks. They also provide mechanism analyses involving numerical sensitivity, fingerprint correlation, and embedding-space geometry.

Significance. If the results are robust, the paper presents a conceptually interesting unification of molecular property prediction, generation, and conformation tasks through a single SMILES-based decoder trained on structured physicochemical knowledge triples. The idea of using a meta-language to avoid irrelevant natural-language grammar is worth exploring, and the authors have committed to releasing code and model weights. However, the current evidence is not yet convincing: the evaluation does not account for possible overlap between the pretraining corpus and benchmark test sets, all benchmark numbers appear to come from single runs without error bars or significance tests, and the docking methodology is insufficiently described. These issues directly affect the central claim of universal proficiency, so the manuscript needs substantial revision before the claimed significance can be accepted.

major comments (5)
  1. [§4.4 and §2.2] The pretraining corpus is stated as "all the molecules available in PubChem" (~110M SMILES), while the evaluation benchmarks (MoleculeNet, AGBT, GPCR, CASF-2016) are standard collections of drug-like molecules. The paper does not report whether any of these benchmark test molecules were removed from the pretraining corpus. Because the pretraining objective includes reconstructing RDKit property values for each SMILES, an overlapping test molecule would allow the model to retrieve memorized associations rather than generalize to unseen chemistry. This directly threatens the paper's core claim of "universal" proficiency across the four task families. Please provide a canonical-SMILES (or InChIKey) overlap analysis between the pretraining corpus and each benchmark's test split, and either remove overlapping molecules or demonstrate that all reported results are unchanged when restricted to non-overlapping molecules.
  2. [§2.2 and Figures 2–4] All quantitative results appear to be based on single runs, with no error bars, confidence intervals, or significance tests. For example, Figure 3(c) reports macro AUROC and the text claims SOTA results in 7 of 9 classification datasets, but without variance across seeds it is impossible to tell whether the differences from baselines are meaningful. Please report means and standard deviations (or full distributions) across at least three to five random seeds for all main benchmark results, and state the number of seeds used. This is needed to support the "outperforms" and "SOTA" claims made throughout the paper.
  3. [§2.3 and §3] The paper's own analysis undermines the universality of the numeric-conditioning mechanism. In Section 2.3, conditional generation for Chi3v and Kappa1 achieves Pearson correlations of only 0.297 and 0.558, and Ipc shows 12.9% (generation) and 29.5% (prediction) %difference. Section 3 explicitly states that "the natural relationships between the numerical values is lost" and notes that 1 and 10 differ by only one character in the discretized token space. These failures are not peripheral: they concern the central premise that discretized <S,P,O> triples preserve continuous numerical semantics. The authors should either scope the universality claim to properties for which the representation is effective, or provide evidence that the benchmark properties (QED, LogP, MolWt, TPSA, etc.) do not exhibit the same failure modes under the exact protocol used in Figure 2.
  4. [§4.3 and §2.2 (binding conformation prediction)] The manuscript does not describe how MolMetaLM performs blind docking on CASF-2016. Section 4.3 explains the translation-rotation-invariant local-frame representation for atomic coordinates, but it does not specify how the model is conditioned on the protein structure, how candidate poses are generated, or how ligand RMSD is computed. Without this methodology, the docking curves in Figure 3(b) are not reproducible, and the claim of proficiency in "conformation inference" is not verifiable. Please add a precise description of the inference procedure used for the CASF-2016 evaluation.
  5. [§4.4 and §2.2] The fine-tuning protocol for downstream tasks is missing. The paper provides pretraining hyperparameters (1,000,000 steps, batch size 256, learning rate 1e-4) but does not state the learning rate, number of epochs, batch size, early stopping, or random seeds used when fine-tuning for property prediction, GPCR activity, or any other downstream benchmark. Please include these details in the manuscript (or explicitly point to a documented configuration in the released repository) so that the central results can be reproduced and compared fairly.
minor comments (6)
  1. [Figure 3(e) caption] The caption says "x-axis indicates 7 molecular property regression datasets," but Section 2.2 says the experiments are on six regression datasets [24]. Please reconcile the number of datasets.
  2. [Throughout] There are several typos and wording errors, e.g., "Firugre" (Figure 3(e) caption), "Benefiting form" (Section 2.1), "long-chian" (Section 2.3), "MACS" instead of MACCS (Section 2.3), and "valid ration" (Section 2.1). A careful proofreading pass is needed.
  3. [§2.1 and Figure 2(b)] The successful ratio for multiple-condition generation is defined in the text as "the ratio of unique generated molecules satisfying the Lipinski's five rules," but the exact ranges for MolWt, NHD, NHA, LogP, and NRB are given only in the surrounding text. Please restate the precise acceptance criteria in the figure caption or in Methods for clarity.
  4. [§4.3] The paper states that the length of all fingerprints is set to 176. Since MACCS keys are usually 166 bits, please clarify whether the fingerprints were folded or truncated and how the 176-bit length was chosen.
  5. [§3] The abstract and introduction claim proficiency in "conformation inference," but the docking result in Section 2.2 is qualified as superior only "when the ligand RMSD cutoff is over 1.4Å." The broader claims should carry the same qualification to avoid overstatement.
  6. [§3] The statement that the meta-language paradigm "spawns tens of thousands of pre-training tasks" is not quantified. Please specify the exact count of distinct task configurations generated from the 18 basic denoising tasks, 402 properties, five fingerprints, and conformation tasks, or use a more modest claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the central benchmark results are external and the pretraining-to-evaluation chain is not definitionally tied to its outputs; the main risks are data overlap and post hoc interpretation, which are not circularity.

full rationale

The paper's derivation chain is: define <S,P,O> meta-language triples (Eq. 1, Section 1), instantiate token/sequence/order denoising tasks (Section 4.2), pretrain a Llama-style decoder on 110M PubChem SMILES with RDKit-computed properties (Section 4.4), then evaluate on external MoleculeNet, AGBT, GPCR, and CASF-2016 benchmarks or on RDKit-scored generated molecules (Section 2). No step rewrites its own output as an input: property values in pretraining are RDKit-derived and benchmark labels come from independent datasets; the model is not fitted to the test labels it is asked to predict. The paper invokes no author-specific uniqueness theorem and no load-bearing self-citation. Its internal admissions in Section 2.3 and Section 3 that 'the natural relationships between the numerical values is lost' and that MolMetaLM 'does not fully understand the continuous relationships between the values which are discretized into tokens' are honest limitations, weakening the universal-proficiency claim but not making it circular. The analogical-reasoning account in Figure 5(c) is post hoc interpretation. The absence of a reported overlap filter between PubChem pretraining and the benchmark sets is a plausible memorization/leakage threat to generalization, but it is a data-hygiene concern requiring overlap analysis, not a reduction of the claimed results to their inputs by construction.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central claim rests on domain assumptions about descriptor sufficiency, tokenized numeric semantics, and conformation encoding, plus standard transformer machinery. The main free design choices are pretraining steps, fingerprint length, and the 402-descriptor property set. No hidden physical entities are introduced.

free parameters (4)
  • Pretraining steps = 1,000,000
    Chosen by hand; no convergence curve or early stopping reported. The final representation quality could depend on this.
  • Fingerprint length = 176
    All five fingerprints are truncated to 176 bits, a design choice that discards fingerprint bits and affects the pretraining task suite.
  • Property set = 402 RDKit descriptors
    The set of physicochemical properties is taken from rdkit.Chem.Descriptors; this defines the vocabulary of object values and bounds the knowledge coverage.
  • Model scale = 12 layers, 12 heads, hidden 768, FFN 2560
    Standard Llama-style hyperparameters with no ablation study showing sensitivity or optimality.
assumptions (4)
  • domain assumption RDKit-computed physicochemical descriptors are a sufficient, unbiased source of molecular knowledge for pretraining.
    All object values in the meta language come from RDKit descriptors (Section 4.4). If these descriptors miss important chemistry, the pretraining tasks inherit that bias.
  • domain assumption Discretized numeric property values retain enough semantic information for conditional generation and prediction.
    The model tokenizes numeric values; the Discussion admits this loses the natural relationship between numbers. Section 2.3 shows failures on Chi3v, Kappa1, and Ipc.
  • domain assumption Local-coordinate representation (alpha, beta, d) from preceding three atoms determines molecular conformations with translation and rotation invariance.
    Section 4.3 assumes this sequential representation is sufficient to reconstruct 3D coordinates and poses; no proof or ablation is given.
  • domain assumption A single decoder-only Llama-style model can learn all molecular tasks through denoising pretraining.
    The framework relies on autoregressive denoising as a universal task interface (Sections 4.1 and 4.2); no comparison against encoder-decoder alternatives is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MolMetaLM: a Physicochemical Knowledge-Guided Molecular Meta Language Model." pith.science (2026). https://pith.science/paper/OCVV67ZA

@misc{pith2026241115500,
  author       = {Pith},
  title        = {Pith review of: MolMetaLM: a Physicochemical Knowledge-Guided Molecular Meta Language Model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OCVV67ZA}},
  note         = {Machine review of arXiv:2411.15500}
}
read the original abstract

Most current molecular language models transfer the masked language model or image-text generation model from natural language processing to molecular field. However, molecules are not solely characterized by atom/bond symbols; they encapsulate important physical/chemical properties. Moreover, normal language models bring grammar rules that are irrelevant for understanding molecules. In this study, we propose a novel physicochemical knowledge-guided molecular meta language framework MolMetaLM. We design a molecule-specialized meta language paradigm, formatted as multiple <S,P,O> (subject, predicate, object) knowledge triples sharing the same S (i.e., molecule) to enhance learning the semantic relationships between physicochemical knowledge and molecules. By introducing different molecular knowledge and noises, the meta language paradigm generates tens of thousands of pretraining tasks. By recovering the token/sequence/order-level noises, MolMetaLM exhibits proficiency in large-scale benchmark evaluations involving property prediction, molecule generation, conformation inference, and molecular optimization. Through MolMetaLM, we offer a new insight for designing language models.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 39 canonical work pages

  1. [1]

    In: Pro- ceedings of NAACL-HLT, pp

    Kenton, J.D.M.-W.C., Toutanova, L.K.: Bert: Pre-training of deep bidirectional trans- formers for language understanding. In: Pro- ceedings of NAACL-HLT, pp. 4171–4186 (2019)

    work page 2019

  2. [2]

    Nature Machine Intelligence 3(12), 1033–1039 (2021)

    M´ endez-Lucio, O., Ahmad, M., Rio-Chanona, E.A., Wegner, J.K.: A geometric deep learn- ing approach to predict binding conforma- tions of bioactive molecules. Nature Machine Intelligence 3(12), 1033–1039 (2021)

    work page 2021

  3. [3]

    arXiv preprint arXiv:2010.09885 (2020)

    Chithrananda, S., Grand, G., Ramsundar, B.: Chemberta: large-scale self-supervised pre- training for molecular property prediction. arXiv preprint arXiv:2010.09885 (2020)

    arXiv 2020

  4. [4]

    In: The Eleventh Inter- national Conference on Learning Representa- tions (2022)

    Zhou, G., Gao, Z., Ding, Q., Zheng, H., Xu, H., Wei, Z., Zhang, L., Ke, G.: Uni- mol: A universal 3d molecular representation learning framework. In: The Eleventh Inter- national Conference on Learning Representa- tions (2022)

    work page 2022

  5. [5]

    OpenAI blog 1(8), 9 (2019)

    Radford, A., Wu, J., Child, R., Luan, D., Amodei, D., Sutskever, I., et al.: Language models are unsupervised multitask learners. OpenAI blog 1(8), 9 (2019)

    2019

  6. [6]

    In: 2022 Conference on Empirical Methods in Natural Language Processing, EMNLP 2022 (2022)

    Edwards, C., Lai, T., Ros, K., Honke, G., Cho, K., Ji, H.: Translation between molecules and natural language. In: 2022 Conference on Empirical Methods in Natural Language Processing, EMNLP 2022 (2022)

    work page 2022

  7. [7]

    arXiv preprint arXiv:2402.06852 (2024)

    Zhang, D., Liu, W., Tan, Q., Chen, J., Yan, H., Yan, Y., Li, J., Huang, W., Yue, X., Zhou, D., et al.: Chemllm: A chem- ical large language model. arXiv preprint arXiv:2402.06852 (2024)

    arXiv 2024

  8. [8]

    In: Proceedings of the 60th Annual Meeting of the Association for Computa- tional Linguistics (Volume 1: Long Papers), pp

    Du, Z., Qian, Y., Liu, X., Ding, M., Qiu, J., Yang, Z., Tang, J.: Glm: General language model pretraining with autoregressive blank infilling. In: Proceedings of the 60th Annual Meeting of the Association for Computa- tional Linguistics (Volume 1: Long Papers), pp. 320–335 (2022)

    work page 2022

Show all 53 references
  1. [9]

    arXiv preprint arXiv:2302.13971 (2023)

    Touvron, H., Lavril, T., Izacard, G., Mar- tinet, X., Lachaux, M.-A., Lacroix, T., Rozi` ere, B., Goyal, N., Hambro, E., Azhar, F., et al.: Llama: Open and efficient foun- dation language models. arXiv preprint arXiv:2302.13971 (2023)

    2023 arXiv

  2. [10]

    Advances in neural information processing systems 35, 27730– 27744 (2022)

    Ouyang, L., Wu, J., Jiang, X., Almeida, D., Wainwright, C., Mishkin, P., Zhang, C., Agarwal, S., Slama, K., Ray, A., et al.: Train- ing language models to follow instructions with human feedback. Advances in neural information processing systems 35, 27730– 27744 (2022)

    2022

  3. [11]

    arXiv preprint arXiv:2306.06615 (2023) 13

    Li, J., Liu, Y., Fan, W., Wei, X.-Y., Liu, H., Tang, J., Li, Q.: Empowering molecule dis- covery for molecule-caption translation with large language models: A chatgpt perspec- tive. arXiv preprint arXiv:2306.06615 (2023) 13

    2023 arXiv

  4. [12]

    Akad´ emiai Nyomda, ??? (1976)

    Jakobson, R.: Metalanguage as a Linguistic Problem. Akad´ emiai Nyomda, ??? (1976)

    1976

  5. [13]

    SEMANTiCS (Posters, Demos, SuCCESS) 48(1-4), 2 (2016)

    Ehrlinger, L., W¨ oß, W.: Towards a definition of knowledge graphs. SEMANTiCS (Posters, Demos, SuCCESS) 48(1-4), 2 (2016)

    2016

  6. [14]

    In: Proceedings of the 58th Annual Meet- ing of the Association for Computational Linguistics, pp

    Lewis, M., Liu, Y., Goyal, N., Ghazvininejad, M., Mohamed, A., Levy, O., Stoyanov, V., Zettlemoyer, L.: Bart: Denoising sequence- to-sequence pre-training for natural language generation, translation, and comprehension. In: Proceedings of the 58th Annual Meet- ing of the Assoc...

    2020

  7. [15]

    In: The Eleventh International Conference on Learning Representations (2022)

    Tay, Y., Dehghani, M., Tran, V.Q., Gar- cia, X., Wei, J., Wang, X., Chung, H.W., Bahri, D., Schuster, T., Zheng, S., et al.: Ul2: Unifying language learning paradigms. In: The Eleventh International Conference on Learning Representations (2022)

    2022

  8. [16]

    IEEE/ACM Transactions on Audio, Speech, and Language Processing 29, 3504–3514 (2021)

    Cui, Y., Che, W., Liu, T., Qin, B., Yang, Z.: Pre-training with whole word masking for chinese bert. IEEE/ACM Transactions on Audio, Speech, and Language Processing 29, 3504–3514 (2021)

    2021

  9. [17]

    In: Findings of the Association for Computa- tional Linguistics: EMNLP 2020, pp

    Cui, Y., Che, W., Liu, T., Qin, B., Wang, S., Hu, G.: Revisiting pre-trained models for chinese natural language processing. In: Findings of the Association for Computa- tional Linguistics: EMNLP 2020, pp. 657–668 (2020)

    2020

  10. [18]

    In: Proceedings of the 2022 5th International Conference on Algorithms, Computing and Artificial Intelli- gence, pp

    Luo, C., Chen, Z., Jiang, X., Yang, S.: Gap sentences generation with textrank for chi- nese text summarization. In: Proceedings of the 2022 5th International Conference on Algorithms, Computing and Artificial Intelli- gence, pp. 1–5 (2022)

    2022

  11. [19]

    Advances in neural informa- tion processing systems 32 (2019)

    Yang, Z., Dai, Z., Yang, Y., Carbonell, J., Salakhutdinov, R.R., Le, Q.V.: Xlnet: Gener- alized autoregressive pretraining for language understanding. Advances in neural informa- tion processing systems 32 (2019)

    2019

  12. [20]

    Weininger, D.: Smiles, a chemical language and information system. 1. introduction to methodology and encoding rules. Journal of chemical information and computer sciences 28(1), 31–36 (1988)

    1988

  13. [21]

    Advanced drug delivery reviews 64, 4–17 (2012)

    Lipinski, C.A., Lombardo, F., Dominy, B.W., Feeney, P.J.: Experimental and computa- tional approaches to estimate solubility and permeability in drug discovery and devel- opment settings. Advanced drug delivery reviews 64, 4–17 (2012)

    2012

  14. [22]

    In: International Confer- ence on Machine Learning, pp

    Jin, W., Barzilay, R., Jaakkola, T.: Junction tree variational autoencoder for molecular graph generation. In: International Confer- ence on Machine Learning, pp. 2323–2332 (2018). PMLR

    2018

  15. [23]

    In: Interna- tional Conference on Learning Representa- tions (2018)

    Jin, W., Yang, K., Barzilay, R., Jaakkola, T.: Learning multimodal graph-to-graph transla- tion for molecule optimization. In: Interna- tional Conference on Learning Representa- tions (2018)

    2018

  16. [24]

    Nature com- munications 12(1), 3521 (2021)

    Chen, D., Gao, K., Nguyen, D.D., Chen, X., Jiang, Y., Wei, G.-W., Pan, F.: Algebraic graph-assisted bidirectional transformers for molecular property prediction. Nature com- munications 12(1), 3521 (2021)

    2021

  17. [25]

    Chemical science 9(2), 513–530 (2018)

    Wu, Z., Ramsundar, B., Feinberg, E.N., Gomes, J., Geniesse, C., Pappu, A.S., Leswing, K., Pande, V.: Moleculenet: a benchmark for molecular machine learning. Chemical science 9(2), 513–530 (2018)

    2018

  18. [26]

    Journal of chemical information and modeling 58(2), 520–531 (2018)

    Wu, K., Wei, G.-W.: Quantitative toxicity prediction using topology based multitask deep neural networks. Journal of chemical information and modeling 58(2), 520–531 (2018)

    2018

  19. [27]

    Journal of chemical information and modeling 47(6), 2140–2148 (2007)

    Cheng, T., Zhao, Y., Li, X., Lin, F., Xu, Y., Zhang, X., Li, Y., Wang, R., Lai, L.: Compu- tation of octanol- water partition coefficients by guiding an additive model with knowl- edge. Journal of chemical information and modeling 47(6), 2140–2148 (2007)

    2007

  20. [28]

    Journal of computational chemistry 31(2), 455–461 14 (2010)

    Trott, O., Olson, A.J.: Autodock vina: improving the speed and accuracy of dock- ing with a new scoring function, efficient optimization, and multithreading. Journal of computational chemistry 31(2), 455–461 14 (2010)

    2010

  21. [29]

    0: New docking methods, expanded force field, and python bindings

    Eberhardt, J., Santos-Martins, D., Tillack, A.F., Forli, S.: Autodock vina 1.2. 0: New docking methods, expanded force field, and python bindings. Journal of chemical information and modeling 61(8), 3891–3898 (2021)

    2021

  22. [30]

    PloS one 11(5), 0155183 (2016)

    Quiroga, R., Villarreal, M.A.: Vinardo: A scoring function based on autodock vina improves scoring, docking, and virtual screen- ing. PloS one 11(5), 0155183 (2016)

    2016

  23. [31]

    Bravais, A.: Analyse Math´ ematique sur les Probabilit´ es des Erreurs de Situation D’un point. Impr. Royale, ??? (1844)

  24. [32]

    Journal of Cheminformatics 12(1), 56 (2020)

    David, L., Thakkar, A., Mercado, R., Engkvist, O.: Molecular representations in ai- driven drug discovery: a review and practical guide. Journal of Cheminformatics 12(1), 56 (2020)

    2020

  25. [33]

    In: Proceedings of the AAAI Conference on Artificial Intelligence, vol

    Fu, Z., Lam, W., So, A.M.-C., Shi, B.: A the- oretical analysis of the repetition problem in text generation. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 35, pp. 12848–12856 (2021)

    2021

  26. [34]

    Chemosphere 179, 49–56 (2017)

    Nolte, T.M., Peijnenburg, W.J., Hendriks, A.J., Meent, D.: Quantitative structure- activity relationships for green algae growth inhibition by polymer particles. Chemosphere 179, 49–56 (2017)

    2017

  27. [35]

    Journal of machine learning research 9(11) (2008)

    Maaten, L., Hinton, G.: Visualizing data using t-sne. Journal of machine learning research 9(11) (2008)

    2008

  28. [36]

    Reviews in computational chemistry, 367–422 (1991)

    Hall, L.H., Kier, L.B.: The molecular connec- tivity chi indexes and kappa shape indexes in structure-property modeling. Reviews in computational chemistry, 367–422 (1991)

    1991

  29. [37]

    Journal of Molecular Graphics and Mod- elling 18(4-5), 464–477 (2000)

    Labute, P.: A widely applicable set of descrip- tors. Journal of Molecular Graphics and Mod- elling 18(4-5), 464–477 (2000)

    2000

  30. [38]

    In: The Eleventh International Conference on Learning Repre- sentations (2022)

    Zhang, N., Li, L., Chen, X., Liang, X., Deng, S., Chen, H.: Multimodal analogical reason- ing over knowledge graphs. In: The Eleventh International Conference on Learning Repre- sentations (2022)

    2022

  31. [39]

    arXiv preprint arXiv:2401.11817 (2024)

    Xu, Z., Jain, S., Kankanhalli, M.: Hallu- cination is inevitable: An innate limitation of large language models. arXiv preprint arXiv:2401.11817 (2024)

    2024 arXiv

  32. [40]

    Advances in neural information processing systems 33, 1877–1901 (2020)

    Brown, T., Mann, B., Ryder, N., Subbiah, M., Kaplan, J.D., Dhariwal, P., Neelakan- tan, A., Shyam, P., Sastry, G., Askell, A., et al.: Language models are few-shot learners. Advances in neural information processing systems 33, 1877–1901 (2020)

    2020

  33. [41]

    In: The Eleventh International Conference on Learning Repre- sentations (2022)

    Zeng, A., Liu, X., Du, Z., Wang, Z., Lai, H., Ding, M., Yang, Z., Xu, Y., Zheng, W., Xia, X., et al.: Glm-130b: An open bilin- gual pre-trained model. In: The Eleventh International Conference on Learning Repre- sentations (2022)

    2022

  34. [42]

    Advances in Neural Information Processing Systems 36 (2024)

    Ge, Y., Hua, W., Mei, K., Tan, J., Xu, S., Li, Z., Zhang, Y., et al.: Openagi: When llm meets domain experts. Advances in Neural Information Processing Systems 36 (2024)

    2024

  35. [43]

    Advances in Neural Information Processing Systems 32 (2019)

    Zhang, B., Sennrich, R.: Root mean square layer normalization. Advances in Neural Information Processing Systems 32 (2019)

    2019

  36. [44]

    arXiv preprint arXiv:2002.05202 (2020)

    Shazeer, N.: Glu variants improve trans- former. arXiv preprint arXiv:2002.05202 (2020)

    2020 arXiv

  37. [45]

    Neurocom- puting 568, 127063 (2024)

    Su, J., Ahmed, M., Lu, Y., Pan, S., Bo, W., Liu, Y.: Roformer: Enhanced transformer with rotary position embedding. Neurocom- puting 568, 127063 (2024)

    2024

  38. [46]

    arXiv preprint arXiv:1710.05941 (2017)

    Ramachandran, P., Zoph, B., Le, Q.V.: Searching for activation functions. arXiv preprint arXiv:1710.05941 (2017)

    2017 arXiv

  39. [47]

    Journal of chemi- cal information and computer sciences 42(6), 1273–1280 (2002)

    Durant, J.L., Leland, B.A., Henry, D.R., Nourse, J.G.: Reoptimization of mdl keys for use in drug discovery. Journal of chemi- cal information and computer sciences 42(6), 1273–1280 (2002)

    2002

  40. [48]

    comparison with other descriptors

    Nilakantan, R., Bauman, N., Dixon, J.S., Venkataraghavan, R.: Topological torsion: a 15 new molecular descriptor for sar applications. comparison with other descriptors. Journal of Chemical Information and Computer Sci- ences 27(2), 82–85 (1987)

    1987

  41. [49]

    Journal of chemical information and modeling 50(5), 742–754 (2010)

    Rogers, D., Hahn, M.: Extended-connectivity fingerprints. Journal of chemical information and modeling 50(5), 742–754 (2010)

    2010

  42. [50]

    Journal of chemical information and modeling 46(5), 1924–1936 (2006)

    Gedeck, P., Rohde, B., Bartels, C.: Qsar- how good is it in practice? comparison of descriptor sets on an unbiased cross section of corporate data sets. Journal of chemical information and modeling 46(5), 1924–1936 (2006)

    2006

  43. [51]

    Nucleic acids research 37(suppl 2), 623–633 (2009)

    Wang, Y., Xiao, J., Suzek, T.O., Zhang, J., Wang, J., Bryant, S.H.: Pubchem: a public information system for analyzing bioactivi- ties of small molecules. Nucleic acids research 37(suppl 2), 623–633 (2009)

    2009

  44. [52]

    Greg Landrum 8(31.10), 5281 (2013)

    Landrum, G., et al.: Rdkit: A software suite for cheminformatics, computational chem- istry, and predictive modeling. Greg Landrum 8(31.10), 5281 (2013)

    2013

  45. [53]

    arXiv preprint arXiv:1910.03771 (2019) 16

    Wolf, T., Debut, L., Sanh, V., Chaumond, J., Delangue, C., Moi, A., Cistac, P., Rault, T., Louf, R., Funtowicz, M., et al.: Hug- gingface’s transformers: State-of-the-art nat- ural language processing. arXiv preprint arXiv:1910.03771 (2019) 16

    2019 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.